Hi Maxime,
I have also noticed the wrong calculation here, thanks for the timely help.
Thanks,
Cheng
> -----Original Message-----
> From: Maxime Coquelin <maxime.coque...@redhat.com>
> Sent: Wednesday, July 7, 2021 6:30 PM
> To: dev@dpdk.org; Jiang, Cheng1 <cheng1.ji...@intel.com>; Xia, Chenbo
> <chenbo....@intel.com>
> Cc: Maxime Coquelin <maxime.coque...@redhat.com>; sta...@dpdk.org
> Subject: [PATCH v2] vhost: fix assuming packed ring size is a power of 2
>
> Unlike split ring, packed ring does not mandate the ring size to be a power of
> 2. So we have to use a modulo operation when wrapping ring index.
>
> Fixes: 873e8dad6f49 ("vhost: support packed ring in async datapath")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Maxime Coquelin <maxime.coque...@redhat.com>
> ---
>
> V2: Also fix wrapping in virtio_dev_rx_async_get_info_idx which is used in
> the packed ring data path.
>
> lib/vhost/virtio_net.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c index
> b93482587c..0de5231db7 100644
> --- a/lib/vhost/virtio_net.c
> +++ b/lib/vhost/virtio_net.c
> @@ -1470,7 +1470,7 @@ virtio_dev_rx_async_get_info_idx(uint16_t
> pkts_idx,
> uint16_t vq_size, uint16_t n_inflight) {
> return pkts_idx > n_inflight ? (pkts_idx - n_inflight) :
> - (vq_size - n_inflight + pkts_idx) & (vq_size - 1);
> + (vq_size - n_inflight + pkts_idx) % vq_size;
> }
>
> static __rte_always_inline void
> @@ -2131,7 +2131,7 @@ uint16_t rte_vhost_poll_enqueue_completed(int
> vid, uint16_t queue_id,
>
> if (vq_is_packed(dev)) {
> for (i = 0; i < n_pkts_put; i++) {
> - from = (start_idx + i) & (vq_size - 1);
> + from = (start_idx + i) % vq_size;
> n_buffers += pkts_info[from].nr_buffers;
> pkts[i] = pkts_info[from].mbuf;
> }
> --
> 2.31.1
Acked-by: Cheng Jiang <cheng1.ji...@intel.com>
